Social patterns of miscarriage reporting and risk: Insights from survey data in France

According to clinical studies, miscarriages, that is spontaneous pregnancy losses before 22–24 weeks of gestation, affect approximately 13 to 18 % of recognized pregnancies. However, clinical studies are rarely representative of the population. Surveys tend to be, but they are subject to misreporting of miscarriages, as suggested by existing work in the US. Thus, existing evidence of social inequalities in miscarriage risk is inconsistent. This study first applies a model to estimate the extent of miscarriage misreporting in the French FECOND survey (2010-11). Estimated 92% of miscarriages were reported and underreporting was slightly more pronounced among women with lower education. Nevertheless, the estimated prevalence of miscarriages (14% of all pregnancies) remained largely unaffected and only with small educational differences. These findings indicate that estimating miscarriage risk factors can be done without correcting for reporting bias in this survey. Second, we used multilevel multinomial logistic regressions to estimate socio-demographic characteristics and epidemiological factors associated with miscarriage. As in previous work, advanced reproductive ages were associated with a higher miscarriage risk. Our models reveal that the association between miscarriage risk and prior reproductive history is explained by unobserved time-consistent individual characteristics, such as potentially. Conversely, education and self-assessed financial conditions were not associated with miscarriage risk. Younger cohorts exhibit a higher miscarriage risk, suggesting either an age effect on recall, a cohort effect regarding pregnancy recognition, or reduced stigma resulting in more reliable reporting of miscarriage in surveys.